How FISA Feeds Fabulists

Kerry Howley | August 7, 2008

Mathematician John Allen Paulos on party games and FISA:

In [Daniel] Dennett's more interesting and suggestive game, one person, the subject, is selected from a group of people at a party and asked to leave the room. He is told that in his absence one of the other partygoers will relate a recent dream to the other party attendees. The person selected then returns to the party and, through a sequence of Yes or No questions about the dream, attempts to accomplish two things: reconstruct the dream and identify whose dream it was.

The punch line is that no one has related any dream. The individual partygoers are instructed to respond either Yes or No to the subject's questions according to some completely arbitrary rule. Any rule will do, however, and may be supplemented by a non-contradiction clause so that no answer directly contradicts an earlier one. The Yes or No requirement can be loosened as well to allow for vagueness and evasion.

The result is that the subject, impelled by his own obsessions, often constructs an outlandish and obscene dream in response to the random answers he elicits. He may think he knows whose dream it is, but then the ruse is revealed to him and he is told that the dream really has no author. In a strong sense, however, the subject himself is the dream weaver. His preoccupations dictated his questions which, even if answered negatively at first, frequently received a positive response in a later formulation to a different partygoer. These positive responses were then pursued.

What's that got to do with warrantless surveillance? Faced with random stores of information, investigators will feel driven to tease out a narrative. The wider the net, the more fertile the ground for storytelling: Delusion will be rewarded, biases confirmed.

Having so much unfiltered information (phone records, emails, internet searches, travel itineraries, financial statements, Facebook postings, credit card bills, etcetera) with which to work and having no requirement for specific focus or judicial warrant, any wild hunch or obsession can be pursued relentlessly without fear of disconfirmation.

The resulting combinatorial explosion of connections and interconnections will always provide ample raw material for the development of any investigative group's pet theory. If, for example, there are 400,000 Americans (and a million names) on the terrorist watch list, as the ACLU announced last month, then there are about 80 billion pairs of possible co-conspirators on the list and more than 10 quadrillion possible threesomes on the list. Search this large "party" diligently enough for confirmatoryYesses and ignore willfully enough disconfirmatory Noes, and who knows what will result.

Via Matt Steinglass, who has more.

Comments

|8.7.08 @ 2:28PM|#

Figures often beguile me, particularly when I have the arranging of them myself; in which case the remark attributed to Disraeli would often apply with justice and force: 'There are three kinds of lies: lies, damned lies, and statistics.'

|8.7.08 @ 2:29PM|#

The BBC show Darren Brown's Trick or Treat performed similar expiriments on people to demonstrate the lengths the human mind will go to to establish patterns in data or evidence. Itr was fascinating to see it in practice.

This is also addressed quite well in Cory Doctorow's Little Brother.

Elemenope|8.7.08 @ 2:31PM|#

Human beings, shock of shocks, are obsessive pattern recognition filters.

Not to be too down on religion generally (which has much to speak for it), but mythologies invariably associated with them are almost completely transparent attempts to assign a narrative of causation to a paucity of observational data.

Why did Johnny get smote by the lightning bolt? He was a bad monkey.

George|8.7.08 @ 2:37PM|#

I find mescaline makes it so much easier for me to understand vast quantities of seemingly unrelated data.

NeonCat|8.7.08 @ 2:42PM|#

I remember watching an interesting program within the last year, either on PBS or History, about the Soviet government around 1983 becoming convinced through accumlated evidence and suppositions that the US was about to launch a first strike on it, all based on little pieces of information and preconceived notions.

It wouldn't have been so bad, but the Soviets were considering launching first - they were so sure they were right.

Episiarch|8.7.08 @ 2:46PM|#

When you have a shitload of data in a database, if you know your SQL, you can actually pull data out pretty easily that has some sort of coherence to it. What it really depends on is what the data is, how good the data integrity is, and if connections have any validity.

|8.7.08 @ 2:48PM|#

I find mescaline makes it so much easier for me to understand vast quantities of seemingly unrelated data.

I find it interesting that on more than one occasion I've heard somebody mispronounce mesclun as mescaline. It brings up a new and disturbing mental image for "tossed salad".

|8.7.08 @ 2:55PM|#

Paulos' argument ignores how police and counter-intelligence investigations actually work and could just as well be used against standard procedures. If there's an accident and witnesses report a red Chevy sedan driven by a white guy fleeing the scene, they don't arrest every white guy who owns a red Chevy sedan. They start with that info, and use other methods to narrow down the suspect list. The watch list is just a place to start, not anything final.

Of course there is "fear of disconfirmation": how many people want to risk their careers by fingering innocents?

I'm not saying the watch list can't be abused, but so far it seems to have been rather free of abuses.

A.G. Pym|8.7.08 @ 3:17PM|#

I remember an episode (perchance the final one?) of the CBS comedy "The Governor and J.J.", in which the Governor's staff received a stack of books and, under the impression that they were materials to be reviewed for a determination of whether they were obscene or not, found lots and lots of "obscenity" in them. The gag was, that this was a stack of history books and research materials that the library had delivered by mistake, including the one joke I recall in particular - one of the books was a history of WWII tanks or some such and, when the Governor asks his chief of staff how it could possibly be obscene, is told that sometimes the tank captain rode above the driver and directed him by tapping the driver's shoulders with his feet. "How is that obscene?" "Foot fetish!"

Plus ca change . . . .

|8.7.08 @ 3:18PM|#

Man, PapayaSF, what alternate universe do you live in? Don't you read Balko's posts where LEOs routinely finger innocents, always killing their dogs and sometimes killing the person, and are never punished? Hell, sometimes they get medals.

|8.7.08 @ 4:05PM|#

Sure, mistakes happen, but few of those are intentional. And "routinely"? In the sense that it happens, OK. In the sense that it represents a more than a tiny percentage of all arrests, no.

thoreau|8.7.08 @ 4:28PM|#

I find that whenever I need a needle it's best to first add as much hay to the stack as possible. Or something.

bill|8.7.08 @ 4:40PM|#

It always amazes me that there are people out there that are willing to carry out such activity. There is no need for Hitler clones. It seems you can always find people to be the Brownshirts, the SS and the Gestapo.

|8.7.08 @ 4:48PM|#

People will often quickly accept inadequate evidence that appears to confirm what they already believe.

Seward|8.7.08 @ 4:51PM|#

PapayaSF,

Sure, mistakes happen, but few of those are intentional.

Intentionality doesn't seem to matter (especially in light of what is in the write-up.) Indeed, couldn't a truly good faith effort be in part the source of the problem?

And "routinely"? In the sense that it happens, OK. In the sense that it represents a more than a tiny percentage of all arrests, no.

Can you provide us some information which demonstrates that it is only a tiny percentage of cases?

|8.7.08 @ 4:53PM|#

"""Of course there is "fear of disconfirmation": how many people want to risk their careers by fingering innocents? """

I'm laughing my ass off on that statement. We have a problem with cops killing innocent people and not much happening to them. Why would finger pointing be worse? The only career they are risking is the one to which the finger points, not their own.

|8.7.08 @ 4:57PM|#

The resulting combinatorial explosion of connections and interconnections will always provide ample raw material for the development of any investigative group's pet theory. If, for example, there are 400,000 Americans (and a million names) on the terrorist watch list, as the ACLU announced last month, then there are about 80 billion pairs of possible co-conspirators on the list and more than 10 quadrillion possible threesomes on the list. Search this large "party" diligently enough for confirmatoryYesses and ignore willfully enough disconfirmatory Noes, and who knows what will result.

The point about it being easy to find spurious correlations is certainly true.

However, at the same time it's currently impossible as well to search through all the possibilities as well. I agree that that sort of large scale data-mining would be frightening-- but if it were possible, it could also just as easily be done with information of public record that requires no FISA, no warrants, no anything to obtain because it's all public knowledge. (Trawl Google and Facebook for starters, among other places.) Claims about the capabilities of the intelligence community tend to be ridiculous.

The point is correct that the government could not possibly do that level of data-mining and actually separate the gold from the dross. The cries about how the FBI should have found the 9/11 hijackers because of their flight training ignored the difficulty of such data searches as well.

|8.7.08 @ 5:00PM|#

When you have a shitload of data in a database, if you know your SQL, you can actually pull data out pretty easily that has some sort of coherence to it. What it really depends on is what the data is, how good the data integrity is, and if connections have any validity.

Some sort of coherence, perhaps, but as you say, if you want validity you have to classify all your BLOBs according to all the possible types of data and connections you'd like to make. That's considerably harder than setting up a database where you already know exactly what to look for and what pieces of information are relevant.

|8.7.08 @ 5:20PM|#

10 quadrillion possible threesomes

The mind boggles.

|8.7.08 @ 6:48PM|#

Seward, I'm going by the number of law enforcement outrages I read about here and elsewhere (perhaps 5 per month?) compared to whatever the total number of arrests and convictions is, which I can't find offhand but must be in the 6 or 7 figures per year. Even if US police are murdering and/or framing 100 people per month (which seems rather high, no?), that would be a tiny fraction of all cases.

TrickyVic, I think that's largely hyperbole on your part. Are there some trigger-happy cops? No doubt. And some innocents get shot. But a mistaken shooting isn't always "murder," and the cops are almost always investigated, tried, and/or disciplined in some way.

Seward|8.7.08 @ 7:13PM|#

PapayaSF,

Seward, I'm going by the number of law enforcement outrages I read about here and elsewhere (perhaps 5 per month?)...

I'm not quite sure why what is reported in the press ought to be viewed as a good representation of the total number. Can you give me a good reason why it should be viewed as a good representation?

...(which seems rather high, no?)...

I have no way to judge if that is a high estimate or a low estimate. For instance, historically it might be a low number or a high one. Given your six or seven figure estimate though it certainly seems well within the realm of possibility.

...that would be a tiny fraction of all cases.

If there were indeed over a thousand cases of framing, etc. each year in the U.S. then even if it were a "tiny fraction" I would find it rather troubling.

|8.8.08 @ 12:36PM|#

It just seems rather paranoid to me to assume that the number of instances of cops framing innocents is anything more than a small fractional percentage of all arrests and convictions. Oh, if you ask people in prison, no doubt some huge percentage say they're innocent, but I've never seen any evidence that the true percentage is much more than the few cases you read about, plus some amount that you don't. I'm not say that small fraction isn't worrisome and unjust, just that I see no reason to think it's anything but a small fraction.

Of course I'm just estimating, but I think it's based on safe assumptions, like the fact that it's often very difficult to convict people of crimes they obviously have committed, which means it's even harder to frame people for crimes they haven't committed.